{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,15]],"date-time":"2026-07-15T12:05:24Z","timestamp":1784117124632,"version":"3.55.0"},"reference-count":22,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,7,15]],"date-time":"2026-07-15T00:00:00Z","timestamp":1784073600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Med."],"abstract":"<jats:sec>\n                    <jats:title>Introduction<\/jats:title>\n                    <jats:p>Automated, objective assessment of rehabilitation movement quality from three-dimensional skeleton recordings can support scalable and low-cost physiotherapy monitoring, but reported performance is often affected by subject leakage, inconsistent class-imbalance handling and non-comparable evaluation protocols.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>This study establishes a leakage-controlled, subject-independent benchmark on the publicly available IntelliRehabDS corpus. After removing 12 ambiguous-label files and 2 sequences shorter than 16 frames, 2,575 sequences from 29 adult and adolescent participants were retained, comprising 2,047 correct and 528 incorrect executions across nine rehabilitation gestures. Sequences were root-centred, torso-scaled, smoothed with a Savitzky\u2013Golay filter, resampled to 64 frames and represented through kinematic features or skeleton tensors. Five classical feature-based learners and the topology-aware ST-SkelNet sequence model were evaluated through nested subject-wise cross-validation. Scaling and SMOTE were fitted only within the relevant training partitions for the classical models, whereas ST-SkelNet used class-weighted binary cross-entropy.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>Performance was reported using fold-averaged metrics, pooled out-of-fold confusion matrices and ROC-AUC analysis. No single model dominated: the multilayer perceptron achieved the highest accuracy (0.869), the random forest the highest ROC-AUC (0.861), and the support vector machine the highest incorrect-class recall (0.629). ST-SkelNet achieved 0.846 accuracy, 0.855 ROC-AUC and 0.612 incorrect-class recall, but did not consistently exceed the strongest classical baselines. Head-motion and bilateral-asymmetry features showed descriptive separation between correct and incorrect executions in this adult\/adolescent cohort.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Discussion<\/jats:title>\n                    <jats:p>The contribution is a reproducible methodological benchmark rather than a clinical system or a state-of-the-art architecture. IntelliRehabDS is treated here as a surrogate adult\/adolescent dataset for methodological validation and benchmarking, not as infant data. Transfer to paediatric populations, including congenital muscular torticollis, requires independently collected age-specific data and prospective clinical validation.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fmed.2026.1878438","type":"journal-article","created":{"date-parts":[[2026,7,15]],"date-time":"2026-07-15T11:10:12Z","timestamp":1784113812000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Skeleton-based rehabilitation movement quality classification: a leakage-controlled benchmark on IntelliRehabDS"],"prefix":"10.3389","volume":"13","author":[{"given":"Zhen","family":"Zhu","sequence":"first","affiliation":[{"name":"Department of Orthopedics, Jiujiang Hospital of Traditional Chinese Medicine","place":["Jiujiang, Jiangxi, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yuqiong","family":"Xiang","sequence":"additional","affiliation":[{"name":"School of Rehabilitation Medicine, Gannan Medical University","place":["Ganzhou, Jiangxi, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xianzhu","family":"Tian","sequence":"additional","affiliation":[{"name":"Department of Orthopedics, Jiujiang Hospital of Traditional Chinese Medicine","place":["Jiujiang, Jiangxi, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,7,15]]},"reference":[{"key":"ref1","doi-asserted-by":"publisher","first-page":"106835","DOI":"10.1016\/j.compbiomed.2023.106835","article-title":"Artificial intelligence for skeleton-based physical rehabilitation action evaluation: a systematic review","volume":"158","author":"Sardari","year":"2023","journal-title":"Comput Biol Med"},{"key":"ref2","doi-asserted-by":"publisher","first-page":"410","DOI":"10.1109\/TNSRE.2022.3150392","article-title":"Graph convolutional networks for assessment of physical rehabilitation exercises","volume":"30","author":"Deb","year":"2022","journal-title":"IEEE Trans Neural Syst Rehabil Eng"},{"key":"ref3","doi-asserted-by":"publisher","first-page":"1586","DOI":"10.3390\/s25051586","article-title":"Machine learning-based computer vision for depth camera-based physiotherapy movement assessment: a systematic review","volume":"25","author":"Zhou","year":"2025","journal-title":"Sensors (Basel)"},{"key":"ref4","doi-asserted-by":"publisher","first-page":"2","DOI":"10.3390\/data3010002","article-title":"A data set of human body movements for physical rehabilitation exercises","volume":"3","author":"Vakanski","year":"2018","journal-title":"Data"},{"key":"ref5","doi-asserted-by":"publisher","first-page":"46","DOI":"10.3390\/data6050046","article-title":"IntelliRehabDS (IRDS)\u2014a dataset of physical rehabilitation movements","volume":"6","author":"Miron","year":"2021","journal-title":"Data"},{"key":"ref6","doi-asserted-by":"publisher","first-page":"370","DOI":"10.1097\/PEP.0000000000001114","article-title":"Physical therapy management of congenital muscular torticollis: a 2024 evidence-based clinical practice guideline from the American Physical Therapy Association academy of Pediatric physical therapy","volume":"36","author":"Sargent","year":"2024","journal-title":"Pediatr Phys Ther"},{"key":"ref7","doi-asserted-by":"publisher","first-page":"69572","DOI":"10.7759\/cureus.69572","article-title":"Effect of pediatric physical therapy interventions on congenital muscular torticollis: a systematic review","volume":"16","author":"Desai","year":"2024","journal-title":"Cureus"},{"key":"ref8","doi-asserted-by":"publisher","first-page":"4111","DOI":"10.1038\/s41598-022-08157-5","article-title":"An efficient self-attention network for skeleton-based action recognition","volume":"12","author":"Qin","year":"2022","journal-title":"Sci Rep"},{"key":"ref9","doi-asserted-by":"publisher","first-page":"107420","DOI":"10.1016\/j.compbiomed.2023.107420","article-title":"D-stgcnt: a dense spatio-temporal graph conv-gru network based on transformer for assessment of patient physical rehabilitation","volume":"165","author":"Mourchid","year":"2023","journal-title":"Comput Biol Med"},{"key":"ref10","doi-asserted-by":"publisher","first-page":"26720","DOI":"10.1038\/s41598-025-12020-8","article-title":"Frame topology fusion-based hierarchical graph convolution for automatic assessment of physical rehabilitation exercises","volume":"15","author":"Zhang","year":"2025","journal-title":"Sci Rep"},{"key":"ref11","first-page":"169","author":"Marusic","year":"2024"},{"key":"ref12","doi-asserted-by":"publisher","first-page":"633","DOI":"10.1038\/s41597-023-02375-w","article-title":"KINECAL: a dataset for falls-risk assessment and balance impairment analysis","volume":"10","author":"Maudsley-Barton","year":"2023","journal-title":"Sci Data"},{"key":"ref13","first-page":"2507.21018","article-title":"Deep learning for skeleton based human motion rehabilitation assessment: a benchmark","volume-title":"arXiv","author":"Ismail-Fawaz","year":"2025"},{"key":"ref14","doi-asserted-by":"publisher","first-page":"3790","DOI":"10.1109\/TNSRE.2023.3317411","article-title":"A contrastive learning network for performance metric and assessment of physical rehabilitation exercises","volume":"31","author":"Yao","year":"2023","journal-title":"IEEE Trans Neural Syst Rehabil Eng"},{"key":"ref15","doi-asserted-by":"publisher","first-page":"225","DOI":"10.3390\/children10020225","article-title":"Quantitative analysis of the head tilt using three-dimensional temporal scan in children with torticollis","volume":"10","author":"Yun","year":"2023","journal-title":"Children (Basel)"},{"key":"ref16","doi-asserted-by":"publisher","first-page":"185","DOI":"10.1021\/acsmeasuresciau.1c00054","article-title":"Why and how Savitzky\u2013Golay filters should be replaced","volume":"2","author":"Schmid","year":"2022","journal-title":"ACS Meas Sci Au"},{"key":"ref17","doi-asserted-by":"publisher","first-page":"0100","DOI":"10.34133\/cbsystems.0100","article-title":"A survey on 3D skeleton-based action recognition using learning method","volume":"5","author":"Ren","year":"2024","journal-title":"Cyborg Bionic Syst"},{"key":"ref18","volume-title":"Hands-on Machine Learning with Scikit-Learn, Keras, and TensorFlow","author":"G\u00e9ron","year":"2022"},{"key":"ref19","doi-asserted-by":"publisher","first-page":"863","DOI":"10.1613\/jair.1.11192","article-title":"SMOTE for learning from imbalanced data: progress and challenges, marking the 15-year anniversary","volume":"61","author":"Fern\u00e1ndez","year":"2018","journal-title":"J Artif Intell Res"},{"key":"ref20","doi-asserted-by":"publisher","first-page":"e49023","DOI":"10.2196\/49023","article-title":"Practical considerations and applied examples of cross-validation for model development and evaluation in health care: tutorial","volume":"2","author":"Wilimitis","year":"2023","journal-title":"JMIR AI"},{"key":"ref21","doi-asserted-by":"publisher","first-page":"2735","DOI":"10.3390\/s22072735","article-title":"Validity and reliability of Kinect v2 for quantifying upper body kinematics during seated reaching","volume":"22","author":"Faity","year":"2022","journal-title":"Sensors"},{"key":"ref22","first-page":"1","author":"R\u00e9by","year":"2023"}],"container-title":["Frontiers in Medicine"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fmed.2026.1878438\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,15]],"date-time":"2026-07-15T11:10:13Z","timestamp":1784113813000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fmed.2026.1878438\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,15]]},"references-count":22,"alternative-id":["10.3389\/fmed.2026.1878438"],"URL":"https:\/\/doi.org\/10.3389\/fmed.2026.1878438","relation":{},"ISSN":["2296-858X"],"issn-type":[{"value":"2296-858X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,15]]},"article-number":"1878438"}}